Data storage in cloud computing

DATA STORAGE.Introduction to enterprise data storage.Data storage management.File Systems.Cloud file system.Cloud data stores.Using Grid for data storage.

1. S.JAMUNA
ASSISTANT PROFESSOR
DEPARTMENT OF COMPUTER APPLICATIONS
BON SECOURS COLLEGE FOR WOMEN
DATA STORAGE

2. Introduction to enterprise data storage
Direct Attached Storage (DAS)
Storage Area Network(SAN)
Network Attached Storage(NAS)

Direct Attached Storage (DAS)
Basic storage system
Used for build SAN & NAS
Performance of SAN & NAS depends on DAS

3. Storage Area Network(SAN)
Used for multiple host to connect single storage device
Simultaneous access not permitted- used for clustering
environment
Technologies- Fibre channel FC, iSCSI, AoE
(ATA over ethernet)

Network Attached Storage(NAS)
File level storage
File server
Advantage – sharing a single volume by multiple host

4. Data storage management
Tiered Storage
Data storage management
Storage resource management tools:
 Configuration tools – handle set up of storage resources
 Provisioning tools- define and control access of storage resources
 Measurement tools- analyze performance
Storage management process
 Data storage management tools rely on policies
Three areas in storage management
Change management
Performance and capacity planning
Tiering

5. Data storage challenges
Massive data demand
Performance barrier
Power consumption and cost
Unified Storage
Combination of NAS & SAN- NUS (Network unified
storage)
Accessed by single & multiple hosts
Advantage- reduced cost, supports fiber channel & iSCSI

6. File Systems
Fat file system
Planned for system with very small RAM & small disks
Require less system resources
NTFS
Much simpler than FAT.
System areas customized while using files
Security incorporated
Not apt for small sized disks

7. Cloud file system
Considerations:
Must sustain basic file functionality
Should be an open source
Should be grown up enough
Should be shared
Should be scalable
Honest data protection

8. Ghost file system
Used in AWS (Amazon web services)
High redundant elastic mounted, cost effective, standards
based file system
Provides fully featured scalable and stable cloud file system
Benefits of Ghost file system
Elastic and cost effective
Multi-region redundancy
Highly secure
No administration
Anywhere

9. Features of Ghost file system
Mature elastic file system
All files & metadata can be duplicated
WebDav for standard mounting on any server or client
FTP access
Web interface
File name search
Side loading of files

10. Gluster File system
Is an open source
Distributed file system
Clusters storage devices over network, aggregating disk,
memory resources & managing data as a single unit
Supports client with valid IP address
Attributes – Scalability, performance, high availability, global
name place, etc…

11. Hadoop File System
Distributed file system to run on a commodity hardware
Files are stored in blocks from 64 mb-1024 mb
Blocks distributed across cluster & replicated for fault
tolerance
Xtreem FS:A distributed and replicated file system
Is a distributed , replicated & open source
Allows to mount & access files via WWW
Replicate files to reduce network congestion, latency &
increase data availability

12. Kosmos File System
Gives high performance with availability and reliability
Deployed in c++ using standard system components
Incorporated with Hadoop and Hypertable

Cloud FS
Is a distributed file system to solve problems
Cloud FS is based on Gluster FS & supported by Red hat &
hosted by Fedora

13. Cloud data stores
Is a data repository – data stored as objects
Includes data repositories, flat files.
Types
Relational databases
Object oriented databases
Operational databases
Schema less data stores
Paper files
Data files

14. Distributed data stores
Is like a distributed database
Non- relational databases-searches quickly over a large
multiple nodes
Eg: Google’s big table, Amazon’s dynamo, Window’s azure.
Types of data stores
Big table – is a compressed , high performance, properietary
storage system
Developed in 2004, used in google applications like google
earth, google map, google book search…
Advantage – scalability,better performance control
Bigtable charts 2 random string values- row & column key and
time stamp in to associated random byte array.

15. Other similar S/W:
Apache Accumulo,Apache Cassandra, Hbase, KDI
Dynamo: A Distributed Storage System
Is a proprietary key value structured storage system or a
dispersed data store
Acts as databases & also distributed hash tables
Most powerful relational database
Is a distributed storage system not a relational database
Advantage- responsive,consistent

16. Using Grid for data storage
Grid storage for grid computing
It virtualizes heterogeneous and remotely located components into a single
system
It allows sharing of computing & data resources for multiple work loads &
enable collaboration
Grid computing uses NAS type of storage
To set unique demands- storage for grid must be flexible
Grid Oriented Storage-GOS
Is a dedicated data storage architecture connected directly to a computational grid
Acts as data bank
Successor of NAS
Deals with long distance, heterogeneous & single image file operations
Acts as a file server